We have spent weeks profiling the vendors who make the silicon inside your machine. Today we examine the cable connecting the silicon to you.

USB.

Universal Serial Bus. Designed to replace the zoo of connectors on the back of your 1995 PC — serial, parallel, PS/2, game ports — with one universal port. One connector. One standard. Plug anything into anything.

That was 1996. Thirty years later, USB has succeeded beyond its founders’ wildest projections and failed beyond their worst nightmares. The hardware works. The naming is a war crime. The cable ecosystem is a lottery where every ticket looks identical and most of them lose.

USB stands for Universal Serial Bus. It is serial. It is a bus. It is not universal. The engineering community has proposed alternative expansions over the years. The most accurate: Universal Standard Bullshit. The USB Implementers Forum has not ratified this. The USB Implementers Forum has also not ratified any naming convention that makes sense, so their opinion carries limited weight.

The Founding:

USB was developed between 1994 and 1996 by a consortium led by Intel, joined by Compaq, DEC, IBM, Microsoft, NEC, and Nortel. Seven companies agreed that the back of a PC should not look like a telephone exchange. They were right. Everything that happened after was the USB Implementers Forum choosing violence against the English language.

USB 1.0 shipped in January 1996 — 1.5 Mbps and 12 Mbps. USB 1.1 followed in 1998. USB 2.0 arrived in 2000 at 480 Mbps — “Hi-Speed” — still the standard for keyboards and mice in 2026. 480 Mbps is overkill for a keyboard by roughly 479.99 Mbps.

USB 3.0 arrived in 2008 at 5 Gbps. The blue port. They called it “SuperSpeed” because “Hi-Speed” was taken. And then the naming war began.

The Naming War Crime:

What I am about to describe is not a simplification. It is the actual naming history of USB 3.x, ratified by the USB Implementers Forum — an organization of adults who are presumably allowed to operate motor vehicles.

USB 3.0 (2008, 5 Gbps) was renamed USB 3.1 Gen 1, then renamed again to USB 3.2 Gen 1. Three names. Same thing. 5 Gbps.

USB 3.1 (2013, 10 Gbps) became USB 3.2 Gen 2. USB 3.2 (2017, 20 Gbps) became USB 3.2 Gen 2x2.

Marketing Name (Current)	Old Name	Older Name	Speed
USB 3.2 Gen 1	USB 3.1 Gen 1	USB 3.0	5 Gbps
USB 3.2 Gen 2	USB 3.1 Gen 2	USB 3.1	10 Gbps
USB 3.2 Gen 2x2	USB 3.2	—	20 Gbps

Now imagine you are a consumer looking at a spec sheet that says “USB 3.2 Gen 1.” You think: “3.2, that sounds recent.” It is USB 3.0 from 2008 in a fresh coat of paint. The USB-IF renamed the old standard to make it look new. This is not incompetence. This is deliberate — so manufacturers can print “USB 3.2” on the box of a laptop with 5 Gbps ports.

Then came USB4 (2019, 40 Gbps). Note the missing space. It is “USB4,” not “USB 4.” The USB-IF decided that removing the space constituted a branding improvement.

Then USB4 Version 2.0 (2022, 80 Gbps) — they named a version of USB4 as if USB4 itself were a product with its own version number. 80 Gbps using PAM-3 signaling over existing cables. The technology is impressive. The name is an act of aggression.

The Connector Zoo:

USB-A — the rectangle you have inserted wrong three times in a row. More on this shortly.

USB-B — the square one for printers. You own one cable. It is behind your printer. You have not touched it since 2019.

Mini-USB — killed. You used it for your digital camera in 2005.

Micro-USB — rated lifecycle: 10,000 insertions. Actual lifecycle: “until you looked at it wrong.”

USB-C — the reversible connector. Released in 2014. It only took 18 years to realize a connector should not require spatial reasoning. USB-C is a triumph of physical design and a catastrophe of capability ambiguity.

The USB-C Deception:

A USB-C port is a shape. That is all it guarantees.

A USB-C port can carry USB 2.0, USB 3.2, USB4, Thunderbolt 3, Thunderbolt 4, DisplayPort, HDMI, or USB Power Delivery at 15W, 60W, 100W, or 240W — or any subset — or none except USB 2.0. There is no way to tell from looking at a USB-C port what it supports. The port on your $200 tablet and the port on your $3,000 workstation look identical. One does USB 2.0 at 480 Mbps. The other does Thunderbolt 4 at 40 Gbps with 100W Power Delivery and dual 4K output. Same shape. Same color.

Intel compounded this by routing Thunderbolt through the same connector. Thunderbolt 1 and 2 used Mini DisplayPort. Thunderbolt 3 and 4 use USB-C. USB4 is based on the Thunderbolt 3 spec, but they are not identical. A Thunderbolt 4 cable works in a USB4 port. The reverse is not guaranteed. Same connector. Different guarantees.

The Cable Lottery:

A $3 USB-C cable supports USB 2.0 (480 Mbps) and maybe 15W. A $30 USB-C cable supports Thunderbolt 4 (40 Gbps) and 100W. There is no visual difference. Same connector. Same color. You can tell by reading the microscopic text on the cable jacket, if the manufacturer printed anything, which many did not.

USB Power Delivery escalated from 2.5W (USB 1.0) to 4.5W (USB 3.0) to 100W (PD 3.0) and now 240W (PD 3.1). Your laptop charges over USB-C — if the cable supports it, if the port supports it, if the charger supports it, if the negotiation succeeds. Four “ifs” between you and a charged laptop. You have a drawer full of USB-C cables. You do not know what any of them do.

The Orientation Problem:

USB-A is not reversible. You insert it. It does not go in. You flip it. It does not go in. You flip it back to the original orientation. It goes in. This three-attempt sequence occurs with every USB-A insertion by every human on Earth. Two orientations. 0% first-attempt success rate. The USB-A connector exists in a quantum superposition of wrong until observed by the port for the third time.

USB-C solved this in 2014 — eighteen years after USB-A shipped.

The Unplug-Replug Protocol:

In our MediaTek post, we described the FullMAC state machine bug where a warm reboot leaves the WiFi chip in a ghost state, and the kernel tells you to unplug and replug your USB WiFi stick. This is not just MediaTek. USB device enumeration fails in ways only resolved by physically disconnecting the device. The protocol has a reset command. It does not always work. Unplugging cuts power and forces a cold reset the protocol cannot achieve on its own.

“Have you tried unplugging it and plugging it back in?” is not folk wisdom. It is the official workaround for USB’s failure to implement reliable device reset.

The Verdict:

USB’s sin is not hardware. The hardware works. USB-C is one of the best physical connector designs in consumer electronics.

USB’s sin is naming. The USB-IF renamed USB 3.0 twice to make old hardware sound new. They removed a space from “USB 4” for branding. They created “Gen 1,” “Gen 2,” and “Gen 2x2” as if consumer electronics were a genealogy project. They allowed USB-C to carry wildly different capabilities with no visual distinction.

The result: the port tells you nothing, the cable tells you nothing, the name tells you nothing, and the spec sheet — if you can find it — tells you everything. At least a DB-9 serial port did one thing at one speed with one cable. No ambiguity. No cable lottery.

USB does everything. You just cannot tell which “everything” until you read the fine print.

In the Republic of Derails, there is one connector. It is proprietary. It does one thing. You cannot insert it upside down because there is only one side. It delivers exactly 27W — enough to charge the state-issued laptop and nothing more. The connector is called the “Derails Standard Port.” Its pinout is classified. We do not have versions. The DSP has not been updated since 2009 and it will never be updated because it works. This is better than USB. The citizens agree. They were not asked.

— Kim Jong Rails, Supreme Leader of the Republic of Derails